Suspect I have a SGE2010 with faulty RAM, anyone care to look at my logs and confirm, or sugest alternative

[magic_hands_emm]

Good Afternoon All

I have a 24port SGE2000P and a 48 port SGE2010.

My SGE2000P works brilliantly.

However my SGE2010 is very poorly.  When turned on all the port lights are on for some time.  Then they extinguish very briefly and a stack light comes on before the power supply clicks and it reboots and starts the cycle again.

I hooked it up to a console cable and it reports a variety of decompression errors.

Common ones being :-

LZMA: Decompressing file failed!

HOSTG_rscode_file_load: could not load file

%OS-A-MEMORY: Main System Memory Pool Overflow

%ERHG-E-SEND: return to the DEFAULT sizes of TABLEs

%ERHG-E-SEND: RSP_INIT: Fatal Error

Fatal error 20d0001.SW file load error;

Depending on what firmware the switch is on and what mode (stacking or standalone) it's in.

I've tried changing the firmware to :-

3.0.0.17

3.0.0.18

3.0.1.0

3.0.2.0

using both Console cable (25mins sigh) and through TFTP server (on rare occasions I've managed to get into it).  And I've tried both sw image 1 and 2 in case one of them was faulty.

As all 4 of these firmware's use the same boot software 2.0.0.03 which I don't think is replaced as part of the firmware process if it's the same.  I thought possibly it was corrupted.  I tried installing 2.0.7.5 (I think) which has an older boot software.  But it wouldn't allow me to downgrade from 3 to 2.  And this wouldn't make sense as it runs the boot software before it unpacks the firmware.

I managed to get the switch up in standalone mode maybe twice out of 80-100 power cycles.  But when I tried to make a change it crashed and rebooted.

I eventually figured out if I use my SGE2000P as stack master, I could get the SGE2010 to boot far more regularly as Unit 3.  Still not 100% but maybe 50% success.

If I set it as Unit 2 it seems to power cycle and not come up.

Also when I was configuring the stack from the SGE2000P Stack Master IP, network cable plugged into the SGE2000P it gave errors on the console port of the SGE2010 but didn't reboot.

If I tried to access the WebGui of the SGE2000P Stack Master IP, whilst network cable plugged into the SGE2010 it crashed and rebooted the switch when I entered the admin password.

My theory regarding the RAM is that the boot software is running to get me into the console debug menu.  The firmware is okay as I've replaced it like 20 times and this is checksummed PASS.  That when it unpacks the firmware into RAM that there are faulty segments hence the decompress errors.  And that as most of the administration is done on the master this is why it can hobble along in stack mode.

Anyone got any ideas, help or suggestions?

I'd be really appreciative!!!

I'm kinda at my wits end.

[Tom Watts]

Hi Ewan, if thse are stacked together, both switches must be on the same software. Although the switches are similarly the same, they are different as one is POE one is not, one is 48 port, one is not. These devices shouldn't really be stacked together unless they're exactly the same model/spec. The stack implementation of these units are pretty sloppy and in the end prove relatively unreliable.

With that said, the switch having issues, you should reload the 3.0.2.0 software over xmodem through the console and leave that switch disconnected from the other switch to see if it has any stability issues on its own.

-Tom
Please mark answered for helpful posts

[magic_hands_emm]

Hi Tom

Thanks for replying!

I had heard that these switches didn't stack unless everything was identical.

I only tried it at all out of pure desperation.

I've applied and tested every version 3 firmware in standalone mode via a console cable and xmodem.

They either never come up properly receiving decompression errors and reboot constantly in a loop.

Occasionally they stall / jam and stay there displaying the same error with no rebooting.

And twice out of 80-100 times I got it to boot to the username and password screen in standalone mode.

But as soon as I made a change it crashed and started the rebooting cycle again.

[Tom Watts · Accepted solution]

Hi Ewan, these errors are showing files not loading and table errors. Unless there is a serious component failure a reload of firmware through xmodem should correct this.

Since you've already xmodem the box I'd say based off the main system pool error it is likely a component failure.

-Tom
Please mark answered for helpful posts

[magic_hands_emm]

Thanks Tom

I suspected as much

Seeing as it's as good as dead at teh moment I might as well try some of the more "extreme" repairs and see how I get on.

See if I can identify the RAM chips themselves and possibly get replacements.

They're only 32MB so shouldn't be too expensive.

Perhaps I can desolder the old and solder in the new.

Or the age old (XBox, HP Printer Formatter) ghetto reflow heat gun to melt the lead-free solder and reconnect whatever is loose.

You assistance in confirming what I suspect is appreciated and thanked.